Redo muon ID CaloTemplates in 3X

http://uaf-2.t2.ucsd.edu/~ibloch/MuID/ http://cmsrocstor.fnal.gov/lpc1/cmsroc/ibloch/MuID

New 3_0 templates

Explanation of these plots:

• Axis:
  • x: track momentum
  • y: energy deposition
• Plots:
  • top row: HO
  • middle row: Hcal
  • bottom row: Ecal
    
  • each row has 5 pads - from left to right: Endcap-, transition region-, barrel, transition region+, Endcap+

Muon templates

=== left: 2_0, right: 3_0 ===

Pion templates

=== left: 2_0, right: 3_0 ===

Performance plots of 3_0 and 2_0 templates

Some performance plots are referenced here. They show the integrated fraction of muons and pions (y-axis) present above a cut on the CaloCompatibility (x-axis). These plots are done in 1 GeV momentum bins of the particles. First plot goes from 1-2 GeV, second from 2-3 GeV and so on, upto 60 GeV.

A second batch of plots further down on the referenced pages shows the integrated fraction of muons relative to the fraction of pions (y-axis), an estimate for the "signal to noise" again above a given cut on the CaloCompatibility (x-axis).

Please find the extensive set of plots here:

• For the 3_0 templates tested on different MC samples from the one the templates were determined from (no bias, but earlier CMSSW version)
• For the 3_0 templates tested on the same MC the new (3_0) templates were determined from (small bias)
• For the 2_0 templates

Validation plots of 3_0 templates

Comparison of the caloCompatibility for v2_0 (current standard) and new 3_0 templates. Results for muons and pions in the momentum range from 0 GeV to 2 TeV are shown.

The spikes in the distributions are coming from muons / pions in the very low (below 3 GeV) and very high (above ~1 TeV) momentum particles.

CaloCompatibility in full eta range

CaloCompatibility in Barrel

CaloCompatibility in Endcap

CaloCompatibility in restricted p range, full eta range

Necessary steps

1. Decide which CMSSW version to use
   1. Use 3_3_0
   2. Further feedback from Hcal/Ecal experts needed?
2. Generate muon and pion MC
   1. prepare 3_3_0 config cards (started here: /uscms/home/ibloch/MC_gen_3_3_0/CMSSW_3_3_0/src)
      -> done
   2. prepare 3_3_0 crab cards
      -> done
   3. test cards
      -> done
   4. run the MC
      -> done
3. Analyze MC to generate raw template histos
   1. Redo the muon collection on the fly when doing the templates (removing the segment match requirement and the p/pt cuts)
      -> tested, set up, works
      -> done
   2. Merge RAW histos
      -> tested, set up, works
      -> done
   3. Normalize RAW histos and bring normalized histos to final format
      -> plot_compatibility_histos.C should be the macro. tested, works
      -> done
4. Check performance of new templates
   -> partially done:
   • tested using new (3_0) templates and running the templateMaker's validation histos.
   • Derek is currently producing new validation ntuples and histos using the new templates.
     • re-produce analysis ntuples redoing the muon collection using the new templates
       -> done
     • re-produce the plots based on the new ntuples
     • check distributions - has the forward fake rate improved?
       • if yes: good
       • if not: need to optimize the diagonal cut (tighten)
5. Check if hcal eta normalization still holds -> tested, works reasonably well:
   

   === left: raw hcal energy, right: pathlength corrected hcal energy ===

   
   -> done
6. Further checks?

Detailed info/questions for steps above

Further steps

The HO calorimeter is being equipped with SiPMs. The sensitivity of the HO to muons will be greatly improved.

• currently only small regions in eta/phi are equipped with SiPMs
• does the CaloComp break in those regions? Or does it work assuming a std HPD response?
• if it breaks, should we produce a SiPM and a HPD HO response template and choose according to eta and phi of the muon?

HOSiPM Muon templates (3_1)

=== left: 3_0, right: 3_1 ===

HOSiPM Pion templates (3_1)

=== left: 3_0, right: 3_1 ===

Available Samples

Summary on DBS

• muon samples
• pion samples

Detailed info

-- IngoBloch - 20-Oct-2009